The Stabilization of Horse Ferrihemoglobin to Acid Denaturation by

May 1, 2002 - Effects of traces of n-alcohols on the acid denaturation of horse ... Thomas A Borgese , John Harrington , Joan M Borgese , Ronald L Nag...
0 downloads 0 Views 1MB Size
256

Biochemistry

J. STEINHABDT, B ONA-PASCUAL, S. BEYCIIOK, AND CHIEN HO

The Stabilization of Horse Ferrihemoglobin to Acid Denaturation by Combination with Ligands* J. SI-EINHARDT,~ R. ONA-PASCUAL, S.BEYCHOK,~ AND C

m Ho

From the Department of Chemistry, Massachusetts Institute of Tachndqgy,Cambri&e Received October 22,1962

The increase in resistance to acid denaturation of ferrihemoglobin when bound to cyanide, azide, fluoride, thiocyanate, or nitric oxide has been studied as a function of pH at three temperatures, and the results have been related to earlier work on the stabilizing effect of formate d on the titration with acid of cyanoferrihemoglobm. A reduction in denaturation velof up to 50-fold has been demonstrated with both cyanide and azide at low (0.004 Y) concentrations at 0.02 ionic strength. At 0.3 ionic strength the same ligand concentration producza an approimately 250-300-fold effect. Formate and fluoride give lesser effects (the latter only at Oo), and thiocyanate does not stabilize at all. Stabilization, when present, d’ . ’ at higher temperatures. Similar but smaller effects are found on the velocity of regemration of denatured protein. By making use of measured valuea of the tendency of each l i g d to &sociate from the protein, it has been poeaible to distinguish the effects of various fractions b o d , which depend on pH, from the intrinsic stabilizing effect of each ligand when bound. The increased stabilities found appear to be consistent with an approximately equal reduction in velocity of the timdependent component of denaturation by those strong negatively charged ligands (cyanide and azide) which form lowgpin (covalent) complexes with the protein-heme, independent of their a t y for the protein. Weaker ligands (fluoride and thiocyanate) which form high-spin complexes stabilize the bound forms to a leeser extent if a t all. Nitric oxide, which is electrically neutral, appears to be without effect, although it is probably bonded CDvalently. The increased stability of the covalent fem-complexes (approximately equal to that of ferro-complexes, which do not differ among themselves) may be associated with the absence in the latter of a positive charge in the vicinity of the iron, rather than with the induction of a strong covalent iron-protein bond in place of the “ionic” bond of ferrihemoglobin. An explanation is proposed for this charge effect in terms of differen- in protem mnformation between ferro- and fem-hemoglobin, ferrihemoglobin bound to ligands, and globin. Regardless of the validity of this mechanism, it is shown that heme or the protoporphyrin nucleus plays an easential part in stabilizing the structure of the protein moiety of native hemoglobin.

In the course of a series of investigations of the increase in acid-binding groups that occurs in various compounds of horse hemoglobin when they are denatured by dilute acid (Steinhardt et al., 1962)’ we have shown that the various forms (carbonmonoxyh?moglobin, HbCO; ferrihemoglobin, Hb +; cyanoferrihemoglobin, HbCN; and others) difFer very greatly in respect to the rates a t which they are denatured by dilute acid a t temperatures between 0’ and 25’. For example, ferrihemoglobin denatures in 0.02 M lactate buffers at 0.3 ionic strength about 250 times faster than when 0.01 M cyanide is present in the same solutions (Steinhardt et ul., 1962). The difference in stability between ferrihemoglobin and carbonmonoxyhemoglobin may be even greater, but exact comparisons have not been made since the rate of acid denaturation of very dilute solutions (0.06%) of the latter is extremely sensitive to traces of oxygen (Zaiser and Steinhardt, 1951). Advantage has been taken of the stabilizing effect of cyanide and CO on hemoglobin to avoid ambiguities in differential titrations of native and denatured hemoglobins which would be introduced by partial denaturation of the native protein at low pH during the few seconds required for measuring the pH after mixing (Steinhardt et al.,1962).

* Some of the results described in this paper were reported a t meetings of the American Chemical Society in April, 1960 (Chicago) and in April, 1962 (Washington, D. C.). t Present address: Department of Chemistry, Georgetown University, Washington 7,D. C. 1Present address: Department of Biochemistry, College of Physicians and Surgeons, Columbia University, New York. 1 Work prior to 195: is summarized in Steinhardt and Zaiser (1955).

These remarkable Merencea m stability of the globin moiety which are brought about by changes m the immediate environment of the iron atom in the prosthetic group of this conjugated protem are attested to not only by difFerences in rates of ’. g but also by differences in ratea of loss of solubility a t the isoelectric point and in rates of change in the characteristic light absorption (Soret band)--all of which run closely parallel, not only in denaturation but in regeneration as well. In the belief that such differences may indicate the direct involvement of the iron of heme, or of heme itself, in the stabilization of the folded form (tertiarystructure) of horse hemoglobin, a detailed investigation of the effect of various ligands on the stability of horse hemoglobin,* and the way in which these effects depend on concentration, temperature, pH, and ionic strength, has been made and is, in part, reported here. l3ecause it is much easier to work with ferrihemoglobin derivatives than with ferrohemoglobin or its derivatives, and the data are consequently less ambiguous, the present paper is mainly concerned with the effect of ligands on protein having the iron in the oxidized form. Results with HbCO, HbOz,and other derivatives of ferrohemoglobin, under anaerobic conditions, and with ferrohemoglobin itself, will be reported elsewhere. EXPERIbU3NTAL

All materials, methods, and procedures were the same as those previously d e s c r i i (steinhardt et al., * Keilin and Hartrec (195%. and K e i h (1960) have cited earlier evidence that ligands of the heme proteins such as oxygen, carbon monoxide, and azide, and the state of oxidation of the iron, &e& the Pesistance of these probins to denaturation by acid, alkali, heat, and chloroform.

Vol. 2, No. 2, Mar.-Apr., 1963 1958; Zaiser and Steinhardt, 1951) except as stated elsewhere. The reagents used as ligands were potassium cyanide (Mallinckrodt Analytical Reagent), sodium azide (Fisher Practical or Purified grade), potassium fluoride (General Chemical Anhydrous Reagent), and potassium thiocyanate (Fisher Certified Reagent). Acetate and lactate buffers were prepared by adding KOH (Baker Analytical Reagent) to the acids (reagent grade). For reasons cited later, two different batches of lactic acid were used (Merck and Mallinckrodt). For denaturation rate measurements freshly prepared solutions of protein (final concentration 0.06%) containing the desired ligand were mixed with buffers (lactate or acetate 0.02 M with respect to the anion) to give the desired p H at the time of initiation of the measurements. In some experiments the ligand salt plus an equivalent quantity of HC1 was used instead of the ligand acid alone. The kinetic results obtained over a period of up to 8 years with both acetate and formate buffers with six different protein preparations have consistently agreed. Except where otherwise stated, the protein was thus initially at a p H different from that of the denaturation run (in the extreme case of cyanide at p H near 8) and at a higher concentration of ligand than after mixing. Since we have observed that, with cyanide, combination of protein and ligand or dissociation occurs as a time-dependent bimolecular reaction (particularly slow at O O ) , it cannot be assumed that the fraction combined with ligand at zero time represents the fraction characterizing an equilibrium at the p H of the experiment. However, in the majority of the experiments with cyanide, in which large excesses of ligand were used, essentially all the protein is combined with ligand except at the lowest range of pH. With azide and other ligands equilibrium is reached so rapidly that the reservation mentioned above does not apply even at low concmtration. Even with low cyanide concentrations (7.5 x M), at which the rate of association is slow, less than 10% difference in the velocity measured resulted from varying the procedure by introducing the cyanide with the buffer instead of adding it to the protein initially. However, when the ligand was introduced with the buffer, a very fast component of the reaction, representing denaturation of the free protein, was over before any experimental observations were made. The ionic strength was always brought to 0.02 with lactate or acetate buffers in the experiments with cyanide, or very slightly higher (up to 0.021) with azide, fluoride, and thiocyanate (since a portion of these added anions are free in the pH range of interest). The equilibria between combined and uncombined protein were studied spectrophotometrically (at 406 mp or 419420 m p or both) at a number of low ligand concentrations at each temp2rature. To assure that the equilibrium constants do not depend on the total charge, z, of the protein (i.e., do not depend on pH) measurements of the equilibria were made at two or three pH values. The range of pH that can be covered is unfortunately limited; at pH values below about 4.3 denaturation is fast enough to prevent attainment of equilibrium, and at pH values over 5.5 the protein exists primarily in the familiar undissociated fourheme form which is characterized by a high degree of interaction between the hemes (Coryell, 1938, and Wyman, 1951). 'rhus, at p H 6.36 the equilibrium of ferrihemoglobin with cyanide is characterized by a Hill exponent, n, of 2.8. At pH values below 5.5, however, none of the ligands used gave appreciable evidence of any interaction, i.e., the equilibria all con-

STABILIZATION OF HORSE FEBREUWOGMBIN

257

formed, within the experimental error, to the general reaction: Hb+

+ HA

=

HbA

+ H+

(1)

and also obeyed the Hill equation with values of n very close to unity. This result is consistcnt with the reported dissociation of horse hemoglotin into halfmolecules at acid p H (Gralen, 1939; Gutter et a/., 1957). The values of the equilibrium constant K., cited in this paper are based on equilibrium concentrations approached by both association and disxciation. The assumptions made are that the thermcdynamic activity coefficient of the protein is not affected by binding ligand, and that the activity coefficient of the ligand acid is unity. Attempts were also made in the case of cyanide to determine K,, by msasuring the velocity constants, k l and k?, for asxociation and dissociation reactions. The association .velocity constant was determined experimentally in the presence of a large excess of cyanide as the pseudo first-order velocity constant divided by the concentration of cyanide. T h e dissociation velocity constant was determined, by diluting an initially 10-fold more concentrated solution, a~ the pseudo first-order velocity constant divided by the hydrogen-ion concentration of the buffer. With 0.06% protein (iron equivalent 3.6 x 10-6 M) and 0.0005 M HCN the average half period for association at 0" was 1.6 minutes, independent of pH; the half-period for dissociation, which should depend on pH, was approximately 75 minutes at all p H values a t this temperature. Thus, the +tially more rapid reaction paths (possibly association of Hb + with HCN rather than with the very dilute CN -, and the dissociation of CN- from HbCN) are not those that determine the final equilibrium. The latter may be governed by a slower reaction of H b + with CN - and of HbCN with the very low concentrations of H+. The ratio of the faster rates gave a fairly constant apparent K , of 0.15 at 0" rather than the average true value, 0.81, cited be10w.~ The values of K , ( = [Hb+][HA]/[HbAl[H+l), presumably independent of ionic strength, used throughout this paper are given in Table I. Values in parentheses are taken from Scheler and Jung (1954), who worked with horse ferrihemoglobin at p H near 6, and whose values for azide are consistent with those we have determined at pH 4-5. No use has been made of the earlier values (Coryell et d . , 1937) obtained with bovine ferrihemoglobin at pH 4.77 (partly because they are considerably higher, and because an incorrect value of K H c appears ~ to have been used in computing K